A few weeks ago, an FDA expert panel by a vote of 6-4 decided against the approval of the use of the antibiotic cefquinome in cattle. Unfortunately, I’ve heard through the grapevine that the political appointees at the FDA plan to overrule the expert panel and approve the use of cefquinome. The chairman of the panel is under pressure to alter the panel’s findings, and the FDA has not posted the minutes of the meeting, which is apparently required by law.

About the post title: cefepime, like cefquinome, is what is known as a fourth-generation cephalosporin antibiotic. While cefquinome is not used to treat people, resistance to cefquinome can also confer resistance to cefepime which is a medically important antibiotic.

There are really two issues at play here. First, this is just a really bad idea. Second, the criteria used by the FDA are cracked. Before I get started, much of the information about the panel meeting can be found at an FDA website, and the risk estimation report is also available as a pdf.
OK, the bad idea. As I mentioned resistance to cefquinome is often correlated with resistance to cefepime. One way this happens is that the bacterium (in this case, E. coli, Salmonella, Klebsiella, and other related bacteria) acquires a gene known as an extended-spectrum beta-lactamase (‘ESBL’). Multiple studies have shown that cefepime resistance is far more likely to occur in bacteria with ESBLs. While the ESBL may only provide partial resistance by itself, when combined with other genetic changes, resistance to cefepime can evolve (and has). (an aside: this process has been observed with related antibiotics).

Many ESBLs are found on plasmids, which are mini-chromosomes that can be transferred from bacterium to bacterium. While the transfer rates typically aren’t high enough to cause an increase in resistance alone, any selective pressure that favors those cells with the plasmid with the ESBL, such as antibiotic use, will result in increased resistance. I want to make something clear: a transfer event is not sufficient–given the large effective population sizes of bacteria, selection due to antibiotic use is required for an increase in resistance. This is very basic population genetics.
In the U.S. and abroad, we’ve already observed disease-causing bacteria (mostly E. coli and Klebsiella) that have ESBLs and that are cefepime resistant. In Ireland, it has even spread to Salmonella, which is disconcerting because cefepime isn’t used in agriculture. This spread probably occurred either because the ESBL is effective against other agricultural antibiotics, or cefepime resistance is linked to other other agricultural antibiotics.

There’s something else this process has illustrated, which is that the FDA criteria for approval are very, well, odd. Here’s what the risk assessment is based on (p. 6):

The overall risk estimation is based upon the following theoretical assumptions:

• The proposed use of cefquinome in cattle may cause resistance in Salmonella spp. (and E. coli) present in the bovine intestinal tract, and
• These resistant Salmonella spp. may contaminate the carcass at slaughter and may transfer to humans via food, and
• These resistant Salmonella spp. may cause infections in humans which require treatment with a fourth-generation cephalosporin (cefepime), and the effectiveness of treatment may be compromised.

I really don’t worry about cefepime resistant Salmonella infections. There just aren’t that many serious Salmonella infections. What I worry about are cefepime resistant Klebsiella and E. coli infections, which can cause pneumonia, sepsis, and urinary tract infections (don’t laugh about UTIs; they kill about 9,000 people/year in the U.S.). The entire focus on Salmonella is foolish. When a cow is treated with cefquinome, all of the harmless E. coli living in it (and there are a lot–around 100,000,000,000 per cow) are exposed to the antibiotic too. That means any E. coli that are cefepime resistant will be favored and increase. These E. coli can then enter the human population directly or transfer these resistance genes to other bacteria (for some reason, this point is made in the first assumption and then ignored in much of the report). Yet for some reason the FDA is obsessed with Salmonella. Salmonella is the tip of the iceberg; you would expect to find resistance genes in Salmonella last, simply because E. coli are so much more numerically dominant in cattle.

The role of regulatory criteria is something to keep in mind as the Coalition of the Sane slowly retakes the American government: the definition of the problem often yields a predetermined outcome. It’s a far more sophisticated version of calling the estate tax the “death tax.” Expect more of this subtle maneuvering now that overt manipulation is no longer easy.

I really hope this doesn’t end badly: after doing such good work on enrofloxacin, it would be a shame to lose ground on cephalosporins.